1984 Volume 19 Issue 1 Pages 47-56
An attempt was made to evaluate the difference of lung cancer mortality in the 23 administration districts of Cental Tokyo. The records are classified into 552 sub-populations according to 3 categories: 12 classes of age, sex and 23 districts. In each population assuming the heterogeneity in their individual levels of mortality risk, the number of deaths have negative binomial distribution. From these assumptions, four hierachical models are constructed to assess the factors which are said to affect lung cancer mortality differencials. Factors in these models are the 3 categories noted above, birth cohorts and degree of diesel exhaust emissions in each district. First, as pointed out in many literatures, the importance of birth cohort effect on lung cancer mortality is recognized. Second, the marked difference on cancer mortality among districts is found: the district which has the highest effect differ more than 1.3 times from the lowest. Further more, in these districts a space clustering related to distriet effect is observed. These results seem to indicate an existence of some environmental factors affecting cancer mortality. So finally, as one of those factors, diesel exhaust of 23 districts are added to the model. The results of this model, however, do not show any significant effects of diesel exhaust emissions to lung cancer mortality differencials.